On Fictional Characters as Types

There is a sense in which Batman in Tim Burton’s film Batman and Batman in Christopher Nolan’s

film Batman Begins are the same fictional character. We call them both ‘Batman’ and this is not a

case of accidental homonymy. These two characters share relevant features, and have common

historical roots, since they both belong to a causal path that springs from Bob Kane’s 1939 comic

Batman.

On the other hand, Burton’s Batman and Nolan’s Batman also appear to be different fictional

individuals, since they have a different look and they act in different fictional worlds. Thus, a

suitable spectator of the film Batman Begins should not treat the Batman that she is seeing on the

screen as the same fictional individual that she encounters while watching Burton’s film Batman. If

she did, she might infer that Batman has undergone a facelift, but this is not a correct inference for a

suitable spectator of Batman Begins.

Thus, with respect to Batman in Burton’s film and Batman in Nolan’s film, we have both

intuitions of sameness and intuitions of difference. There is a sense in which they are the same

fictional entity and there is a sense in which they are not.

In order to reconcile these contrasting intuitions, I propose to treat fictional characters as types

the tokens of which are particular individuals in particular fictional worlds (henceforth, I will use

the expression ‛fictional character’ to designate the type and the expression ‛fictional individual’ to

designate the token). On this proposal, Burton’s Batman and Nolan’s Batman are the same fictional

character inasmuch as they both instantiate the BATMAN type (henceforth I will use capital letters

for the name of a fictional character understood as a type), but they are different fictional

individuals inasmuch as they are distinct tokens of this type.

Types allow us to reconcile intuitions of sameness and difference for a variety of entities.

Consider two inscriptions of the word ‘door’. If we treat the word as a type, we can state that (a) the

1

two inscriptions are the same thing inasmuch as they both instantiate the same word, and yet (b)

they are different things inasmuch as they are distinct inscriptions having different spatiotemporal

locations. That is to say that the two inscriptions are the same at the type level but are distinct at the

token level. Likewise, one can say, Burton’s Batman and Nolan’s Batman are the same fictional

character at the type level but they are distinct fictional individuals at the token level.

The claim that fictional characters are types raises three main worries, however. First, types

are usually taken to be abstract forms, eternal and unchangeable. In contrast, fictional characters are

typically construed as historical outcomes of cultural practices – at least if one endorses, as I do, an

‘artefactualist’ or ‘creationist’ conception of fictional characters.1 Let us call this the creation issue.

1 Fred Kroon and Alberto Voltolini, in ‛Fiction’ (The Stanford Encyclopedia of Philosophy, Fall

2011 Edition, Edward N. Zalta (ed.), URL =

<http://plato.stanford.edu/archives/fall2011/entries/fiction/>), characterize the core of

creationism about fictional characters in the following terms: ‘The intuition that story-tellers

have some kind of creative role to play is accounted for by so called artifactualist, or creationist,

accounts of fictional entities (see Searle (1979), Salmon (1998), Thomasson (1999), Voltolini

(2006); the position was also defended in Kripke’s unpublished John Locke Lectures (1973)

[now published in Kripke (2013)], and elements of the position are found in van Inwagen’s

(1979) theory of fictional objects as posits of literary criticism; Ingarden (1931) is a significant

historical forerunner). According to such accounts, fictional objects are artifacts since they come

into being once they are conceived by their authors; to that extent, they are authorial creations’.

The works referred to by Kroon and Voltolini in this quotation are the following: Roman

Ingarden, Das Literarische Kunstwerk (Tübingen: Niemeyer, 1931); Saul Kripke, Reference and

Existence (Oxford: Oxford University Press, 2013); Nathan Salmon, ‛Nonexistence’, Noûs 32, 3

(1998), 277–319; John R. Searle, ‛The Logical Status of Fictional Discourse’, in P.A. French et

al. (eds.), Contemporary Perspectives in the Philosophy of Language (Minneapolis: University

of Minneapolis Press, 1979), 233–243; Amie Thomasson, Fiction and Metaphysics (Cambridge:

2

Secondly, the tokens of a type normally are concrete particulars in the actual world, as for example

the inscriptions of a word, whereas the alleged tokens of a fictional character are concrete

particulars in a fictional world, and therefore, unlike ordinary tokens, lack a spatiotemporal location

in the actual world. Let us call this the instantiation issue. Thirdly, many fictional characters, unlike

Batman, Sherlock Holmes or Hamlet, only appear in one work of fiction, and therefore, since they

have a unique instance, one can wonder whether it makes sense to treat them as types. Let us call

this the uniqueness issue.

The main aim of this paper is to address these issues in order to defend a creationist account of

fictional characters as types, thereby exploiting the (above-mentioned) explanatory advantages

which the notion of type provides us with. In other words, I will try to clarify the claim that fictional

characters are abstract artefacts by arguing that such artefacts belong to the ontological category of

types.

1. The creation issue

The creation issue can be addressed by characterizing the type as a special ontological category.

This is the strategy pursued by Peter Strawson: ‘The general title of ‘types’, often, though rather

waveringly, confined to words and sentences, may well be extended. I have in mind, for example:

works of art, such as musical and literary compositions, and even, in a certain sense, paintings and

works of sculpture; makes of thing, e.g. makes of motor-car, such as the 1957 Cadillac, of which

there are many particular instances but which is itself a non-particular; and more generally other

things of which the instances are made or produced to a certain design, and which, or some of

Cambridge University Press, 1999); Peter Van Inwagen, ‛Creatures of Fiction’, American

Philosophical Quarterly, 14 (1977), 299–308; Alberto Voltolini, How Ficta Follow Fiction. A

Syncretistic Account of Fictional Entities (Dordrecht: Springer, 2006).

3

which, bear what one is strongly inclined to call a proper name, e.g. flags such as the Union Jack.’2

From Strawson’s perspective, a type is a historically established principle of construction of

like particulars, which are called the type’s tokens.3 A type so construed exhibits a peculiar

normativity, which consists in accurately specifying how a token ought to be constructed. Thus,

according to Strawson, terms such as ‘table’ or ‘bed’ do not designate types; they are nothing but

predicates by means of which we can group things that share the same function but can have

unrelated histories of production.4 By contrast, the tokens of a type are all members of a unique

2 Peter Strawson, Individuals (London: Methuen, 1959), 231.

3 Ibid., 233. Linda Wetzel, in her book Types and Tokens: An Essay on Universals (Cambridge:

MIT Press, 2008), proposes an account of types that is more ontologically welcoming than that

of Strawson. According to Wetzel, a type is any abstract entity that (1) can have instances and

(2) satisfies Frege’s and Quine’s requirements for objectuality. Thus, in her account, biological

species such as ‘the Tarahumara frog’ (2008, 11) or physical notions such as ‘the

electromagnetic field’ (2008, 19) also count as types. Strawson’s account of types as principles

of construction of like things is narrower than Wetzel’s since the former requires that a type be

the outcome of some agency. One could rename the types as Strawson conceives of them with

terms such as ‛artefactual types’ or ‛pragmatic types’ or ‛Strawsonian types’. Still, for the sake of

simplicity, in this paper I will simply call them ‛types’.

4 Strawson, Individuals, 233. Ruth Garrett Millikan in On Clear and Confused Ideas: An Essay

about Substance Concepts (New York: Cambridge University Press, 2000) and Crawford Elder

in Real Natures and Familiar Objects (Cambridge: MIT Press, 2004) also emphasize this

distinction. Millikan states that substances such as ‛chair’, unlike substances such as ‛1969

Plymouth Valiant 100’, ‛do not fall under well-focused substance templates’ (2000, 30). Elder

states that ‛Chairs are less likely to compose a copied kind than [...] Eames desk chairs of the

1957 design’ (2004, 149). That being the case, both Millikan’s notion of a well-focused

substance template and Elder’s notion of a copied kind have important affinities with Strawson’s

4

causal chain or tree, which springs from a unique principle of construction that has been established

in certain historical circumstances. A type is precisely this historically created principle of

construction. As Andrew Harrison puts it in his discussion of Strawson’s account, types ‘are not

laid up in Heaven, but in cultures’.5

A Strawsonian type can function as both a principle of collection and a principle of evaluation

of its tokens in virtue of being a principle of construction of these tokens. For example, we can say

notion of a type, but there is also one crucial difference. What matters for falling under a

Millikanian template or an Elderian kind basically is being produced through a copying

mechanism or process. Thus, not only artefactual notions but also biological notions can satisfy

this requirement and therefore can be treated as Millikanian templates or Elderian kinds. By

contrast, Strawsonian types are principles of construction of like tokens that are rooted in the

normativity of a social practice, and can therefore only have artifacts as instances.

5 Andrew Harrison, ‛Works of Art and Other Cultural Objects’, Proceedings of the Aristotelian

Society, 68 (1967), 124. An account of types that is similar to the one Strawson and Harrison

propose can be found in the following works: David Davies, ‛Varying Impressions’, The Journal

of Aesthetics and Art Criticism, 73, 1 (2015), 81–92; ; Robert Howell, ‘Types, Indicated and

Initiated’, The British Journal of Aesthetics, 42, 2 (2002), 105–127; Lee Walters, ‛Repeatable

Artworks as Created Types’, The British Journal of Aesthetics, 53, 4 (2013), 461–477; Richard

Wollheim, Art and its Objects: An Introduction to Aesthetics (New York: Harper and Row; 2nd

edn. revised, 1980). By contrast, other philosophers are more inclined to take the term ‛type’ as a

synonym of ‛universal’ and therefore introduce new expressions in order to designate the

ontological specificity of the Strawsonian type. Consider, for instance: David Kaplan, ‛Words’,

Aristotelian Society Supplementary Volume 64 (1990), 93–119; Guy Rohrbaugh, ‛Artworks as

Historical Individuals’, European Journal of Philosophy, 11, 2 (2003), 177–205; Barry Smith,

‛John Searle: From Speech Acts to Social Reality,’ in Smith, B. (ed.), John Searle (Cambridge:

Cambridge University Press, 2003), 1–33. Kaplan speaks of ‘non-physical objects’ having

5

that a certain inscription is an incorrect instance of a poem since it contains some typos, that is: that

inscription has been constructed according to the standard specified by the type (that is why it

counts as a token) but it does not fully comply with that standard (that is why it counts as an

incorrect token). By conceiving of the type as a principle of construction of like tokens, we can

assess a certain token by considering the degree to which this token complies with the principle that

governs its construction.6

Types can be created in two different ways: either directly as principles of constructions, i.e.

as sets of instructions, or indirectly by means of what Strawson calls ‘a model particular’.7 For

example, a work of music can be created either by writing a score or by means of a paradigmatic

performance (symphonies are usually created in the former way, songs in the latter).

It is worth stressing that the model particular is not itself a type. On the one hand, a type is a

non-particular individual, which has a beginning in time but not a place in space. On the other hand,

the model particular is a full-fledged particular that has a location in both time and space. A type is

a normative principle that specifies features but does not possess those features, whereas the model

particular is a concrete entity that, just like other tokens, actually possesses the features specified by

the type. Thus, the model particular resembles other tokens, whereas there is no resemblance

‘physical embodiment at many places and times’; Smith speaks of ‘an abstract formation with a

beginning, and perhaps an ending, in time’; Rohrbaugh speaks of ‘a structure-preserving and

transmitting mechanism which undergirds the continued life of an historical individual’. Yet, for

the aim of this paper, these are mainly lexical differences, which I will neglect in what follows.

6 In this respect, types are like Wolterstorff’s Platonic ‘norm kinds’. Cf. Nicholas Wolterstorff,

Works and Worlds of Art (Oxford: Clarendon Press, 1980), 56. The view of fictional characters

as normative kinds that Wolterstorff defends in his book has, Platonism apart, some affinities

with the one advocated in this paper.

7 Strawson, Individuals, 233. I owe the acknowledgement of the distinction between two ways of

creating a type to a remark by the anonymous referee.

6

relation between the type and its tokens. This is because a principle of construction does not

resemble the objects the construction of which it governs. As Harrison puts it: ‘The ‛Form’ Imperial

Standard Yard, unlike the paradigm instance, cannot be of a yard in length, not indeed of any

length’.8

With this Strawsonian conception of types in place, we are well positioned to address the

creativity issue for the view that fictional characters are types. Recall that the issue arises because

types are supposed to be abstract and immutable entities, whereas fictional characters are clearly

historically created. If nothing else, this apparent contrast suggests that fictional characters have a

beginning whereas types do not. If so, fictional characters cannot be types. However, Strawson’s

contributions in this area show that there is a legitimate notion of type that does not cast types as by

definition eternal and beginning-less. Fictional characters, understood as Strawsonian types, can be

created.

There is at least one class of cases in which fictional characters can be explicitly created as

principles of construction of like fictional individuals: theatre. A playwright can create a certain

fictional character directly as a principle of construction of like fictional individuals that can be

instantiated by several actors. However, in most forms of fiction, fictional characters are normally

created by means of a model particular. In this case, normativity is acquired by means of

exemplarity: the features that the model particular exhibits are the features that the future instances

ought to have. Even if some features of the model particular may by more important than others, in

principle any feature of the model particular has normative relevance. If some instance lacks a

feature that the model particular has, one is in principle entitled to remark on this flaw and possibly

ask for reasons. It is important to stress that the notion of flaw, here, is not to be interpreted from an

8 Harrison, ‛Works of Art and Other Cultural Objects’, 115. Beyond the debate on types, the locus

classicus for this sort of reflection can be found in Ludwig Wittgenstein, Philosophical

Investigations (London: Blackwell, 1953), § 50: ‘There is one thing of which one can say neither

that it is one metre long, nor that it is not one metre long, and that is the standard metre in Paris’.

7

aesthetic point of view, let alone from a moral point of view. It is just a matter of compliance with

the model particular.

In the case of Batman, the model particular is to be found in issue 27 of Detective Comics,

published in May 1939. In this issue the writer and artist Bob Kane (with the help of the writer Bill

Finger) introduces a new character, ‛the Bat-Man’, which has since functioned as a principle of

construction of like fictional individuals. The latter are the further instances of the type BATMAN

that we can encounter in successive TV series, films or other comics such as Frank Miller’s

Batman: The Dark Knight Returns.

The functioning of a type such as BATMAN can be described, in abstract terms, as follows.

Given a fiction-maker A who introduces a fictional individual x in her work of fiction, if the

relevant cultural practice recognizes x as being constructed by A through the use of a pre-existing

character C as principle of construction (instead of as introducing a brand-new principle of

construction), then x is a new instance of C. Otherwise, x constitutes the original, exemplary token

of a brand-new character X created by A.

I endorse, here, the account of practices proposed by Peter Lamarque: ‛The practices

associated with works of all kinds are constituted by complex networks of beliefs, conventions,

specialist terminology, and basic, often unstated, presuppositions.’9 More specifically, I treat a

cultural practice as a repository of shared information, a sort of implicit encyclopaedia that

corresponds to the common knowledge of a community. From this perspective, the functioning of a

type such as BATMAN can be rephrased as follows. Given a fiction-maker A who introduces a

fictional individual x in her work of fiction, if x is subsumed under a pre-existing encyclopaedia

entry for a character C, then x is a new instance of C. If, instead, a new encyclopaedia entry is

expressly created for x as a character, then x is the model particular of a brand-new character X

created by A.

9 Peter Lamarque, Work and Object: Explorations in the Metaphysics of Art (Oxford: Oxford

University Press, 2010), 10.

8

If this is right, by creating Batman, Kane did something more than just creating a particular

fictional individual. He also implicitly created a principle of construction of like fictional

individuals, and therefore also a standard or rule for evaluating other fictional individuals that are

constructed following that principle. That is, in addition to creating Batman, he created BATMAN.

Suppose now that some children read issue 27 of Detective Comics and then start together a

pretence in which one of them plays the role of Batman. The child who plays this role is supposed

to take into account Batman’s features as established by Kane. As such, she can be assessed by the

other children with respect to the conformity of her impersonation to Kane’s model particular. Thus,

she can be blameworthy for failing to take on a certain feature present in the model particular. One

of the other children might for example blame her by saying: ‛you are wearing a red suit, but the

Bat-Man’s suit should be grey’.

In a similar manner, if a writer imports the character Batman into a novel, or a director imports

it into a play, they are in principle exposed to the same kind of evaluation, and can be found

blameworthy (or praiseworthy). It is important to stress that the notions of correctness and

blameworthiness, here, are not to be interpreted from an aesthetic point of view, let alone from a

moral point of view. It is just a matter of compliance with the common knowledge of a community.

If the character BATMAN has certain features established by its model particular, and one creates a

new instance of BATMAN that lacks some of these features, a member of the community is entitled

to claim that this instance of BATMAN is incorrect inasmuch as it lacks such features. This claim

does not involve either aesthetic or moral blame. It is just a way of signalling that this new instance

deviates from the cultural norm that constitutes the character BATMAN.

For example, if one wrote a work of fiction in which Batman is a dog, one would have the

right to say that this is an irregular token of Batman. Irregular, once again, does not mean

aesthetically wrong or morally wrong. This just means that this instance lacks a feature of the

character BATMAN. In this example, the feature ‘human being’ seems especially relevant, and

9

therefore this instance of BATMAN seems especially irregular. Yet, my account is not committed

to a distinction between essential and accidental features. In my view, as said above, all the features

of the model particular are normative since, in principle, one member of the community can assess a

new instance of a fictional character by pointing to whatever feature of the model particular and

saying ‘the new instance is irregular because it lacks this feature possessed by the model particular.’

Whether the possession of a certain feature of the model particular is much or less relevant for the

degree of correctness of a new instance is an empirical matter that goes beyond the theoretical

approach of this paper.

To sum up, even if Kane, in creating Batman, did not intend to create a principle of

construction of like fictional individuals but only a particular fictional individual, the features he

ascribed to Batman can function as a principle of construction (and of assessment) of like fictional

individuals.10

A trait of fictional characters as types is what may be called the possibility of feedback. This is

the possibility that a new instantiation of a type may end up modifying the normative content of the

type instantiated. If a child plays Batman engaging in pretence with her friends, she is just

instantiating BATMAN without any feedback on the normative content of that type. But if a

cartoonist publishes a successful new comic with Batman as its protagonist, this instantiation of

BATMAN can have a significant feedback on the normative content of the type itself. This is what

happened for example to Miller, whose comics not only instantiated the type BATMAN but also

introduced some relevant new features of it. The features introduced by Miller’s work are now part

of the normative content of the type BATMAN in the disjunctive form ‘if you want to instantiate

BATMAN, you should comply either with Kane’s specifications or with Miller’s specifications’.

For example, Nolan’s movie Batman begins instantiates the BATMAN type by complying with

Miller’s specifications.

10 Thanks to the anonymous referee for drawing my attention to the relative autonomy of the

normativity of the fictional character as a type from the fiction-maker’s intentions.

10

The feedback phenomenon teaches us something important about the identity conditions of the

kind of types fictional characters are. For note that modifying the normative content of a type

through feedback does not amount to creating a new type. The type preserves its identity, even if

new normative features are added. We do not need a new encyclopaedia entry about Batman as a

fictional character in order to take into account Miller’s Batman. We just need to update the

encyclopaedia entry that we created for Kane’s Batman, by taking into account some new features.

The phenomenon of feedback, from this perspective, precisely consists in an update of the

encyclopaedia entry for a certain fictional character. If, instead, a new encyclopaedia entry is

created, then we do no longer have a feedback on an existing character, but the creation of a brand-

new character, which may have been influenced by the former character, but cannot be treated as an

instance of it. 11

11 I am speaking here of an ideal implicit encyclopaedia, which corresponds to the totality of the

shared understandings of a certain cultural practice. However, actual encyclopaedias are often

good approximations of this ideal encyclopaedia. Consider for example some excerpts of the

Wikipedia entry for Batman: ‘Batman is a fictional superhero appearing in American comic

books published by DC Comics. The character was created by artist Bob Kane and writer Bill

Finger, and first appeared in Detective Comics #27 (May 1939). Batman’s secret identity is

Bruce Wayne, an American billionaire, playboy, philanthropist, and owner of Wayne

Enterprises. […] Batman became popular soon after his introduction in 1939 and gained his own

comic book title, Batman, the following year. As the decades went on, differing interpretations

of the character emerged. The late 1960s Batman television series used a camp aesthetic, which

continued to be associated with the character for years after the show ended. Various creators

worked to return the character to his dark roots, culminating in 1986 with The Dark Knight

Returns by Frank Miller. The success of Warner Bros.’ live-action Batman feature films have

helped maintain public interest in the character’. Thanks to the anonymous referee for leading

me to reflect in more depth on the source of the types’ normativity. The notions of ‘ideal

11

As a result of the phenomenon of feedback, fictional characters as types can change over time.

This consequence flows from the Strawsonian account of types as historical non-particular

individuals. The changeability of a type entails that the possibility to determine its normative

content is not an absolute prerogative of the model particular. Through the phenomenon of

feedback, other tokens can play a similar role by establishing new features that will be henceforth

specified by the type as a principle of construction of like tokens. Accordingly, to say that fictional

characters are types is not to commit to their being immutable entities, as the creativity objection

charges. Still, types are relatively stable entities. Changing a type requires a challenging negotiation

within the cultural practice in which the type has been established as a principle of construction of

like tokens. By negotiation, here, I mean an attempt to update an entry of the implicit encyclopaedia

corresponding to the shared understandings of a community. In this respect, Strawsonian types are

norms in force within a cultural practice, and thus resemble juridical entities such as constitutions.12

Just as the latter, the former can be supplemented with new normative contents without losing their

identity. Just as a constitution can be supplemented with new amendments while remaining the

same constitution, so a fictional character (as a type) can be supplemented with new versions (i.e.

new clusters of features that are established by the phenomenon of feedback), while remaining the

same fictional character. And just as constitutions are in practice typically relatively stable, and

amendments relatively rare, so it typically is with fictional characters. The identity of the fictional

character across changes, just as that of a constitution, is warranted by the shared understandings of

encyclopedia’ and ‘feedback’ are the results of such a reflection.

12 Thomasson draws on the analogy between fictional characters and abstract artifacts such as laws

of state; see Fiction and Metaphysics, 130. Likewise, Luke Manning treats fictional characters as

‘constitutive rules’; see his ‛Real Representation of Fictional Objects’, The Journal of Aesthetics

and Art Criticism, 72 (2014), 13–24. The account of fictional characters I propose is sympathetic

with Thomasson’s and Manning’s views but differs from them inasmuch as I claim that the

normativity of fictional characters is connected to their nature of types.

12

the relevant cultural practice.

2. The instantiation issue

The poem-type The Wild Swans at Coole can be instantiated by a token that is an inscription. The

flag-type Union Jack can be instantiated by a token that is a coloured surface. Both inscriptions and

coloured surfaces are concrete particulars in the actual world. But the tokens of a fictional character

such as BATMAN cannot be concrete particulars in the actual world. Certainly, in the actual world,

there are inscriptions and coloured surfaces that support the instantiation of BATMAN, but those

inscriptions and surfaces are not tokens of the type BATMAN. This is because the type in question

is a principle of construction of like fictional individuals, rather than a principle of construction of

like inscriptions or of like coloured surfaces. What the type BATMAN specifies are the features of

a fictional person, and a token that exhibits these features should be a fictional person.

Here is the specificity of fictional characters with respect to ordinary types such as poems or

flags: the tokens of a fictional character cannot be concrete particulars in the actual world. This

specificity raises the issue of the instantiation of fictional characters. What exactly are the tokens of

such types? How can there be a type which is created in the actual world but whose tokens are not

in the actual world? How can we apply a principle of construction of like tokens if the entity to be

constructed is not in the actual world?

I suggest that these questions can be addressed by means of the notion of a mental file,

understood as a mental particular that functions as a vehicle of singular thought, i.e. as ‘a repository

of information that the agent takes to be about a single individual’.13 The instantiation of a fictional

character cannot produce a fictional person in the actual world, but it can nevertheless produce a

mental file in the mind of the suitable recipient of a certain work of fiction. And this file purports to

refer to an individual in the fictional world who should exhibit the features specified by the type as

13 Robin Jeshion, ‛Singular thought: Acquaintance, semantic instrumentalism, and cognitivism’, in

Jeshion, R. (ed.), New Essays on Singular Thought (Oxford: Oxford University Press, 2010) 131.

13

a principle of construction of fictional individuals. Thus, the fictional character actually is

instantiated by a mental file, which allows the suitable recipient of a work of fiction to purport to

refer to the fictional individual that constitutes the token of the fictional character understood as a

type. Fictional characters, so understood, are peculiar types, which lack actual tokens but

nevertheless have actual instances, namely mental files, which purport to refer to the individuals

exhibiting the features specified by the types, namely the purported tokens.

I am drawing here on a subtle distinction between tokens and instances of a type. By ‘token’ I

mean something that exhibits the features specified by the type while by ‘instance’ I mean

something that enables the enjoyment of the features specified by the type. In the case of ordinary

types such as flags or poems, instances and tokens coincide. Conversely, in the case of fictional

characters, there actually are only instances, namely mental files, that allow the enjoyment of the

features specified by the type through a purported reference to an individual that exhibits these

features, namely a purported token. In other words, the fictional character is instantiated by a mental

file that elicits a phenomenology as of reference to a purported token from a suitable recipient of the

work of fiction: ‛The thought fails to have a singular content, though phenomenologically it feels as

if it had a singular content’.14

To sum up, while types such as poems or flags are directly instantiated by their tokens, the

specificity of fictional characters is to be instantiated by actual mental files that point to purported

tokens. The purported tokens exhibit the features specified by the type, but it is in fact the mental

file that enables the enjoyment of these features on the part of a suitable recipient of the work of

14 François Recanati, ‛Mental Files: Replies to my Critics’, Disputatio, 36, (2013), §1. Ultimately,

what makes the instantiation possible is that there is such thing as purporting to refer to a merely

intentional object; that is something that any theory of reference and intentionality must account

for. The reason why I prefer to account for this in terms of mental files is that they trace back

reference and intentionality to mental particulars, which can effectively play the role of

instances.

14

fiction.

The relevant mental files for the process of instantiation of fictional characters belong to the

category that François Recanati calls ‘public files’, that is, ‘files shared by distinct individuals in a

community’ (2012, 205).15 A file that purports to refer to a certain fictional person is public since

the core information to be stored in it is not arbitrarily established by its owner, nor depends on her

particular perspective on the object the file is about. Instead, the relevant information is determined

by a publicly accessible source, namely the work of fiction. In this sense, the file is shared by all

suitable recipients of that work of fiction.

More generally, given a public mental file (MF), I can open my private mental file (MF1) and

you can open your private mental file (MF2). However, both our private files will purport to refer to

the same object, and we will be both committed to storing in our private files the same core

information about that object. This is why MF1 and MF2 can be traced back to a unique public file,

MF, which functions as a norm by which private mental files ought to abide. Just as the private

mental file is the information we take to be about a singular individual, so the public mental file is

the information we ought to take to be about a singular individual (and that we actually take to be

so, if we behave as suitable recipients).

The public file, so understood, can be seen as a development of Geach’s notion of intentional

identity: ‘We have intentional identity when a number of people, or one person on different

occasions, have attitudes with a common focus, whether or not there actually is something at that

focus’.16 However, the notion of a public file (at least as understood in this paper) requires not only

that the attitudes have a common focus, but also that they ascribe the same core information to the

object that is, or is purported to be, at that focus.

15 François Recanati, Mental Files (Oxford: Oxford University Press, 2012).

16 Peter T. Geach, ‘Intentional identity’, The Journal of Philosophy, 64, 20 (1967), 627. I want to

thank the anonymous referee for leading me to clarify the notion of ‘public file’ by linking it to

the notions of ‘intentional identity’ and ‘suitable recipient’.

15

In the case of a fictional individual, the core information is constituted by all the features

unequivocally ascribed to that individual by the work of fiction in which it appears.17 I am assuming

here that within a cultural practice there is significant agreement as to the features that a certain

work of fiction unequivocally ascribes to a certain fictional individual. I believe that this

assumption reflects a basic trait of our cultural practices.18

The notion of a public file leads us to the notion of a suitable recipient of a work of fiction, i.e.

a recipient who stores in her mental files concerning fictional individuals precisely the core

information specified by that work of fiction. Thus, in the case of a work of fiction, the public files

can be conceived of as the mental files deployed by the suitable recipient of that work.

The public mental file enables the instantiation of the fictional character understood as a type,

i.e. as a principle of construction of like fictional individuals. More specifically, the public mental

file generated by a certain work of fiction allows the suitable recipient of this work to purport to

refer to a fictional individual whose features comply with the specification established by the type.

Although there is not a concrete token in the actual world, the fictional character is instantiated by a

mental file that allows the suitable recipient to enjoy a purported token. Ultimately, the public

mental file is the link between the type as a principle of construction that holds in the actual world

and the token as an inhabitant of the fictional world, i.e. the spatiotemporal framework in which the

17 I use the adverb ‘unequivocally’ in order to rule out the features of a fictional character that a

work leaves undetermined or open to a more or less arbitrary interpretation. For example ‘having

been transformed into a vermin’ is an unequivocal feature of the fictional character Gregor

Samsa whereas ‘having been transformed into a beetle’ is not (according to some interpretation

of Kafka’s Metamorphosis, Gregor has been transformed into a cockroach).

18 This is a point that, here, I simply assume, thereby trying to articulate it in terms of types, tokens

and mental files. Arguing for this very point is beyond the scope of this paper. For a thorough

account of the grounding of the works’ normativity in cultural practices’ see Lamarque, Work

and Object.

16

suitable recipient of a work of fiction locates the events of the story told by this work.

The present account allows us to explain an interesting phenomenon regarding the

individuation of fictional individuals, namely, the fact that sometimes the very same individual

occurs at different works of fiction but sometimes different works involve different fictional

individuals that token the same fictional character. A new instantiation of a pre-existing fictional

character occurs when the suitable recipient of fiction is required to open a new mental file instead

of recalling the mental file in which she previously stored information about a token of this

character. For example, the suitable spectator of Gus Van Sant’s 1999 film Psycho should open a

new file about the fictional individual Norman Bates instead of retrieving the file that he or she

originally created while watching Alfred Hitchcock’s 1960 Psycho. Indeed, a spectator who stored

information about Van Sant’s Bates in the same mental file in which she previously stored

information about Hitchcock’s Bates would be completely misunderstanding Van Sant’s Psycho –

arguably, she would enjoy it as a weird story about a man who disguises himself not only as his

mother, but also as another man. That is why Van Sant’s film produces a new token of the type

NORMAN BATES – a new fictional individual numerically distinct from the one Hitchcock’s film

is about. By contrast, a suitable spectator of James Cameron’s 1982 film Aliens should not open a

new mental file about the fictional individual Ellen Ripley, but rather retrieve the file that he or she

originally created while watching Ridley Scott’s 1979 film Alien. In this case, Cameron’s film

Aliens does not create any new token of the type ELLEN RIPLEY, but just exploits a previous

token.

Thus, there are two different ways in which an already existing fictional character can be

imported into a new work of fiction; either by instantiating a new token (the Bates case) or by

exploiting a previous token (the Ripley case). Let us call the former ‘type importation’ and the latter

‘token importation’. This distinction, on which our file model allows us to shed some light, is

usually overlooked in the debate about importation of fictional characters, yet plays a significant

17

role in cultural practices of fiction appreciation. For example, this distinction usually underlies the

distinction between a sequel, in which the token of a certain character is imported (token

importation), and a reboot, in which a new token of a certain character type is instantiated (type

importation).

Interestingly, the feedback (see § 1) of the work of fiction that imports a character on the

fictional character itself is different depending on the kind of importation. On the one hand, in type

importation, the new work of fiction can introduce new features of the fictional character in a

disjunctive manner, inasmuch as the possible successive instances shall take into account either the

old features or the new ones. On the other hand, in token importation, the new work of fiction

introduces new features of the fictional character in a conjunctive manner, inasmuch as the possible

successive instances shall take into account both the old features and the new ones. Token

importation is normally at work in cinematic sequels such as the Alien saga as well as in TV series,

whereas type importation can be found in adaptations, remakes and reboots.

Finally, it is worth noting that the dependence relation between mental files and the

individuals to whom they refer is inverted in the case of fiction with respect to the case of reality. In

the latter, we open a new mental file because we encounter a new real person, whereas in the former

we encounter a new fictional person because we are required to open a new mental file. This

inversion of the dependence relation occurs because fictional individuals, unlike real ones, do not

exist file-independently. Both are intentional objects of corresponding files, but only fictional

individuals are merely-intentional objects, i.e., are nothing but intentional objects. A fictional

individual is nothing but the purported reference of the mental files generated by a certain work of

fiction. When a work of fiction prescribes to create a new mental file for a pre-existing fictional

character C (instead of prescribing to retrieve a file about a former token of C), a new token of C is

produced through the creation of the mental file.

The inversion of the dependence relation between mental files and the individuals they are

18

about helps us to understand why fictional characters, unlike real people, can have multiple

instances. In the case of a real person, the mental file is just a cognitive device that allows us to

track the person, or have ‘an epistemically rewarding relation’ to this person.19 By contrast, in the

case of a fictional character, the mental file is a sort of ontological device, as it were, that leads to

the instantiation of the type. Therefore, if a suitable recipient of fiction creates a new file, a new

instantiation of a fictional character is carried out.

Real individuals, unlike fictional characters, are not types. At most, a type can stem from a

real person if he or she is imported into a work of fiction. In this case, the suitable recipient of

fiction should open a new file in which to store information about this person as an inhabitant of a

fictional world. Thus, the real person imported into fiction can be associated with a peculiar type,

which has one token in the real world and one token in the fictional world. For example, by

importing Glenn Gould in the novel Der Untergeher, Thomas Bernhard created a type GLENN

GOULD, which is ‘tokened’ (scare quotes, as I shall explain in what follows) in the actual world by

a genuine particular on which mental files depend, and is tokened in the fictional world of Der

Untergeher by a purported token, which depends on mental files.

That being the case, the type-token model provides us with a unified account of importation of

both fictional characters and real individuals into works of fiction. This is an important explanatory

advantage of the model, which allows us to clarify what it is to be the ‘fictional surrogate’ of a real

individual.20 With respect to both fictional characters and real individuals, importation amounts to

19 Recanati, Mental Files, 20.

20 I am drawing here on the idea that the importation of a real individual into a work of fiction

amounts to the creation of a new fictional individual, namely a ‘fictional surrogate’ of the real

individual. See Andrea Bonomi, ‘Fictional Contexts’, in P. Bouquet et al. (eds.), Perspectives on

Contexts (Stanford: CSLI Publications, 2008), 215–250; Peter Lamarque and Stein H. Olsen,

Truth, Fiction, and Literature (Oxford: Clarendon Press, 1994), 126 and 293; Alberto Voltolini,

‘Probably the Charterhouse of Parma Does Not Exist, Possibly Not Even That Parma’,

19

the construction of a new purported token through the opening of a new mental file on the part of

the suitable recipient of the work of fiction. But there is also an important difference to be

highlighted. In the importation of a fictional character, the latter already exists as a type having at

least one purported token, and a new purported token is constructed (with a possible feedback on

the type itself). Instead, in the importation of a real individual, it is the importation itself that creates

the fictional character as a type that has a proper (purported) token in the fictional world and an

improper (real) token in the real world. For example, Glenn Gould can be associated with a type,

namely GLENN GOULD, which has a proper token in the fictional world of Der Untergeher and

an improper token in the real world. I call the latter token ‘improper’ since it can be seen as a token

from the perspective of a recipient of the work of fiction, but it is not properly a token. That is

because its existence is ontologically independent from that of the type.

It is worth noting that the token that fits better with the specification of the type is the proper

(purported) token, not the improper (real) one. For example, the token that fits better with the

features of the type GLENN GOULD specified by Der Untergeher is the fictional individual called

‘Glenn Gould’ in the novel, not the real Glenn Gould. From this perspective, the improper (real)

token can be treated as an incorrect or defective instantiation of the type that is correctly

instantiated by the proper (purported) token. Such an incorrectness allows us to estimate the

accuracy of the work of fiction with respect to the real individual who has been imported. The lower

the incorrectness of the improper (real) token with respect to the type, the higher the accuracy of the

proper (purported) token with respect to reality.

3. The uniqueness issue

So far, I have argued that fictional characters such as Batman can be treated as types since they

function as principles of construction of like fictional individuals in distinct fictional worlds.21 But

Humana.Mente, 25, (2013), 235–261.

21 As said above, I understand a fictional world as the spatiotemporal framework in which the

20

what about a fictional character that appears only in one work of fiction, and therefore only in one

fictional world? Let us consider the case of Finn Hudson, a fictional character played by the actor

Cory Monteith in the TV series Glee. When Monteith suddenly died, on July 13, 2013, the author of

Glee, Ryan Murphy, decided that Finn would also die in the fictional world. In this case it seems

that the fictional character Finn Hudson has just one purported token, which is inextricably

connected to Cory Monteith’s performance.

I argue that having just one token does not prevent an entity from being a type. Indeed, a

principle of construction of like things remains a principle of construction of like things even if in

fact it is used only to construct one thing. By adapting a term from artistic printmaking, I will call

‘monotypes’ the fictional characters that in fact have just one token. As David Davies points out, in

artistic printmaking we can find works that in principle might be used to produce several tokens but

are actually used to produce only one token: ‘The status of monotypes as singular artworks might

also be thought to depend upon how they are treated in our artistic practice’.22 Likewise, I argue, the

status of fictional characters as singular entities depends upon how we treat them in our cultural

practices.

A fictional character such as Finn Hudson is a monotype in the sense that it actually has one

token but, in principle, it might have other tokens. For example, one might write a novel in which

Finn Hudson exists in a fictional world different from that of Glee, thereby leading the suitable

reader of this novel to create a new mental file about Finn Hudson. In this way, a new token of the

type FINN HUDSON would be generated.

The monotype character has a uniqueness that makes it resemble a real person, but it also has a

fictionality that makes it resemble fictional characters that are ‘pluritypes’ (i.e. types having several

tokens). Yet, both these resemblances also involve crucial differences. On the one hand, the

monotype character differs from a real person in that the latter enjoys a self-standing existence as a

suitable recipient of a work of fiction locates the events of the story told by this work.

22 Davies, ‛Varying Impressions’, 84.

21

particular individual in the actual world whereas the former is instantiated by a purported token in

the fictional world through a public mental file produced by a work of fiction in the actual world.

On the other hand, the monotype character differs from a pluritype character such as Batman since,

in the actual world, the former seems to exist only as a component of a work of fiction whereas the

latter exhibits a certain independence from any particular work of fiction (as a principle of

construction of like fictional individuals who appear in different works of fiction).

The notion of a template can be used to clarify the distinction between a monotype character

and a pluritype character. I borrow the term ‛template’ from Noël Carroll’s account of films as

types.23 According to Carroll, the film-template is the entity (for example a filmstrip or a sequence

of digits) that allows the film-type to be instantiated in a screening. Likewise, the character-

template, as understood in this paper, allows a fictional character to be instantiated by a mental file

whereby the suitable recipient of a work of fiction can enjoy a purported token. More specifically, I

conceive of a template of a fictional character as the part of a work of fiction that is responsible for

the creation of the public mental file that instantiates that character.24 In other words, a template of a

fictional character is the part of a work of fiction that determines the core information that the

suitable recipient of that work should store in the mental file about the purported token of that

character (see § 2). While the type governs the construction, by fiction-makers, of like fictional

individuals who appear in different works of fiction, the template governs the construction, by the

suitable recipient of a work of fiction, of a certain fictional individual who appears in that work of

23 Noël Carroll, Theorizing the Moving Image (Cambridge: Cambridge University Press, 1996), 67.

Thanks to the anonymous referee for leading me to clarify the functioning of the template in the

case of fictional characters.

24 The relation that holds between the template and the purported token of a fictional character

might be compared to the relation that holds between the genotype and the phenotype of a

biological individual. In both cases, the former element (template/genotype) determines the

appearance of the latter (purported token/phenotype).

22

fiction.

A monotype character has just one template, whereas a pluritype character has different

templates, one in each distinct work of fiction. In other words, a pluritype character generates

distinct purported tokens through distinct templates. In this sense, the monotype entirely owes its

normativity to the work of fiction to which it contributes. By contrast, a pluritype enjoys a

normativity of its own that spreads across several works of fiction.

However, each monotype potentially is a pluritype inasmuch as the features it specifies

through a template in a certain work of fiction might be used to construct other templates as

components of other works of fiction – and therefore other purported tokens. Although the

normative content of the monotype coincides with that of the template through which the monotype

is instantiated, the monotype and the template remain distinct entities. That is because, as noted, the

monotype governs the possible construction (by fiction-makers) of like fictional individuals

whereas the template governs the construction (by the suitable recipient of a work of fiction) of a

certain fictional individual.

A monotype turns into a pluritype when it is actually used to construct a new template as a

component of a distinct work of fiction that leads to the construction of a new purported token.

Batman itself was a monotype when it was originally created by Kane in 1939. It became a

pluritype only when it was used as a principle of construction of a new purported token (arguably in

1943, when the first Batman TV series was released).

The phenomenon of adaptation is often crucial in order to turn a monotype character into a

pluritype. Fictional characters in books or comics can become pluritypes by means of an adaptation

in a film or in a TV series. Symmetrically, a character in a film or in a TV series may become a

pluritype by means of literary adaptations, as nowadays often happens in so-called ‘fan fictions’.25

Finally, it is worth highlighting the distinction between literature and cinema, on the one hand, and

25 A fan fiction is a fiction about characters from an original work of fiction, created by fans of that

work. It has become particularly popular since the advent of the Web.

23

theatre, on the other. In forms of art such as literature or cinema, a fictional character may be

created as a monotype and then become a pluritype. By contrast, in theatre the fictional character

behaves from the beginning as a pluritype inasmuch as it functions as a principle of construction of

like performances for several actors (i.e. as a role).

4. Conclusion

In this paper, I have tried to clarify the claim that fictional characters are abstract artefacts, by

arguing that fictional characters are abstract artefacts in the sense that they are types. On my

account, this means they are principles of construction of like purported tokens, that is, fictional

individuals to whom suitable recipients of fiction purport to refer through public mental files.

The tokening of a fictional character by a fictional individual requires the construction of a

mental file, and this construction is carried out by a work of fiction. More specifically, the

construction of a fictional individual is carried out by the part of a work of fiction that specifies the

features of that individual. I have called this part of the work of fiction a template of the fictional

character.

Usually, when a work of fiction creates a new character, the latter is initially just a monotype,

which corresponds to one template, by means of which the (so far unique) purported token is

generated. This (so far unique) purported token is the model particular of the fictional character.

Still, even in this basic monotypical form, the fictional character remains a type, which, in principle,

might be tokened by a multiplicity of fictional individuals generated by distinct templates that

function as components of distinct works of fiction. When this potentiality is actually exploited, the

fictional character becomes a pluritype whose normativity spreads across a multiplicity of works of

fiction.

In fact, the claim that fictional characters are types is not a brand-new idea. Similar accounts

were proposed, for example, by Nicholas Wolterstorff and Peter Lamarque.26 As Anthony Everett

26 Wolterstorff, Works and Worlds of Art; Lamarque, Work and Object: Explorations in the

24

writes, ‘Peter Lamarque and Nicholas Wolterstorff identify fictional characters with types-of-

people, where we might think of these as being sets of properties or descriptive conditions’.27 Still,

as Everett points out, a view such as Wolterstorff’s or Lamarque’s is a ‛Platonic view’ that finds it

hard to explain the historicity of fictional characters.28 By contrast, in the account of fictional

characters as Strawsonian types that I have proposed, types are conceived of as historical entities,

not as eternal and immutable Platonic Forms. My account explains how such types can be created,

instantiated and possibly modified within our cultural practices, and to that extent takes the type

view a step further.

In his critical analysis of Wolterstorff’s and Lamarque’s accounts, Everett writes: ‘I am

perfectly willing to accept the existence of types-of-people. And I grant that they may sometimes

play an important role in the way we think and talk about fiction. But they are not fictional

characters; they are not what we talk and think about when we think about the character of Holmes

Metaphysics of Art.

27 Anthony Everett, The Nonexistent (Oxford: Oxford University Press, 2013), 143.

28 Everett, The Nonexistent, 141. Both Wolterstorff and Lamarque try to connect the fictional

character as a Platonic Form to the historical circumstances of its creation. Wolterstorff (Works

and Worlds of Art, 144–145) writes ‘a person-kind is not properly called a ‘character’ until some

work has been composed of whose world it is a component’. And Lamarque: ‘As characters per

se, and thus types, they are not created, for sets of properties exist just to the extent that the

properties themselves exist [...]. However, as fictional characters they are created, to the extent

that their grounding narratives are created, and to the extent that they can be individuated in a

quasi-indexical manner by pointing to the source narrative’ (Work and Object: Explorations in

the Metaphysics of Art, 200–201). However, in both accounts, the fictional character remains

essentially tied to an eternal and immutable set of properties, namely a Platonic Form. For a full-

fledged criticism of Wolterstorff’s view along these lines, see Thomasson, Fiction and

Metaphysics, 58–60.

25

or talk about Austen creating Emma’.29 Ultimately, in this paper I have argued that we can conceive

of types in such a way that they really are what we talk and think about when we think about the

character of Holmes or talk about Austen creating Emma.

29 Everett, The Nonexistent, 143, my emphasis. Everett criticizes Lamarque’s and Wolterstorff’s

accounts in the framework of his fictionalist account of fictional characters. According to

fictionalist theorists, fictional characters are nothing but things to which we pretend to refer in

the games of make-believe which works of fiction provide us with. The locus classicus for

fictionalism is Kendall L. Walton, Mimesis as Make-Believe (Cambridge: Harvard University

Press, 1990). Indeed, fictionalism is a version of a more general view called ‘antirealism’,

according to which fictional characters are nothing but objects of thought. For other antirealist

accounts of fictional characters, see: Tim Crane, The Objects of Thought (Oxford: Oxford

University Press, 2013); Stacie Friend, ‘Notions of Nothing’, in M. Garcia-Carpintero and G.

Marti (eds.) Empty Representations (Oxford: Oxford University Press, 2014). I guess that the

account of fictional characters as types that I have proposed can effectively address the

objections that the supporters of fictionalism (and more generally antirealism) raise against the

defenders of realism. But this shall be the topic of another paper.

26